The present application relates to laser diodes, and more particularly to extracting light from a waveguide and coupling that light to a fiber or other device.
Transmission of light through waveguides has been pursued for many types of communications applications. Light signals offer many potential advantages over electronic signals. Light sources are commonly created from semiconductor devices, and include semiconductor devices such as LEDs (Light Emitting Diodes) and LDs (Laser Diodes).
Optical fiber is the most commonly used transmission medium for light signals. A single fiber is capable of carrying several different modulated signals within it at one time. For instance, wavelength division multiplexing divides the used bandwidth of the fiber into different channels (each channel containing a small range of wavelengths) and thus transmits several different wavelengths (or signals) of light at once. Using such a system requires sources for the different wavelengths. More wavelengths on the fiber require more sources to be coupled to the fiber.
Efficient coupling of light into a fiber is simplified if the laser beam has a cross sectional profile that matches the profile of the fiber mode(s). Efficient use of light for communications requires that the light have high temporal coherence. Efficient coupling of light to monomode guides requires spatial coherence. Spatial coherence requires the laser to operate in a single lateral and transverse mode. Temporal coherence requires the laser to operate in a single longitudinal mode and implies a very narrow bandwidth, or range of wavelengths.
The most coherent semiconductor lasers use resonators based on grating feedback rather than Fabry-Perot resonators with reflective end facets. Distributed feedback (DFB) lasers use a Bragg reflective grating covering the entire pumped length of the laser. An alternative to DFB lasers is the use of distributed Bragg reflectors (DBRs) located outside the pumped region.
In conventional DFB and DBR lasers, light is removed through an end facet and the output beams have dimensions entirely controlled by the vertical (i.e., normal to the surface) (x) and lateral (y) size and the composition of the guiding structure. Such output beams are typically have too great a divergence for effective coupling to optical fibers, or for other applications requiring beams with low divergence angles.
Beam dimensions (in at least one direction) larger than that available from laser facets may be obtained by using a Bragg grating to couple light out of the waveguide normal (or at certain fixed angles) to the waveguide surface. So called second order Bragg gratings have a period equal to the wavelength of light of the guided mode. The second grating order of such a grating reflects some of the light back in the waveguide plane while the first order couples some of the light normal to the plane. So called first order (Bragg) gratings have a period equal to one half the wavelength of light in the guided mode, reflect light in the waveguide plane, and do not couple light out of the waveguide. First, second, and third order (etc.) gratings are sometimes referred to as being in resonance. A non-resonant grating couples light out of the waveguide at an angle to the normal and does not reflect any light in the waveguide plane.
U.S. Pat. No. 5,970,081 to Hirayama et al. appears to show a laser with a distributed feedback (DFB) grating of second order or higher that claims to obtain a Gaussian shaped output beam by narrowing the waveguide or using a chirped grating at the outcoupling portion. They do not seem to recognize that by so doing the resonant wavelength of the grating is altered along the length of the narrowing or chirping. This would be expected to result in an output which will fan in angle along the longitudinal direction rather than produce a simple Gaussian intensity variation emitted normal to the plane as claimed. They do not define the beam shape in the lateral direction. In all versions they choose second order outcoupling gratings which, absent a narrowing waveguide or chirp, would emit light perpendicular to the surface of the laser waveguide.
U.S. Pat. No. 4,006,432 to Streifer et al. appears to show a grating out-coupled surface emitting DFB laser. The grating period may be chosen to be either resonant or not.
A paper by Bedford, Luo, and Fallahi titled Bow-Tie Surface-Emitting Lasers (IEEE Photonics Technology Letters, Vol. 12, No. 8, August 2000) appears to show a DBR laser with curved second order grating at the ends to couple light out of the waveguide. The same gratings are used for outcoupling and for reflecting the light within the waveguide. They mention the use of non-resonant gratings in conjunction with reflector gratings if emission at other than the direction normal the waveguide plane is desired. The paper appears to show a flared resonator region which allows symmetric outcoupling from both ends of the laser. This facilitates two outputs that are coherent with one another, with emission in the same direction. Such a device is meant to alleviate the complications of controlling the relative phase between the two emitters.
The Tiberio article (Facetless Bragg reflector surface-emitting AlGaAs/GaAs lasers . . . , J. Vac. Sci. Technol., B9(6), 1991) appears to show a surface emitting laser diode that uses first order reflective gratings and either second order (or non-resonant) gratings for outcoupling. Thus, depending on the chosen grating period, the outcoupled beam can be normal or at an angle to the surface.
U.S. Pat. No. 6,064,783 to Congden appears to show a DBR laser with a grating assisted waveguide coupler that couples light from the laser waveguide to a parallel fiber-like glass waveguide for later coupling to a fiber. Several different lasers are coupled to similar fiber-like glass waveguides in the figures. The fiber axis is parallel to the laser waveguides. This reference mentions that this model is easily attached to a fiber through xe2x80x9cbutt coupling.xe2x80x9d The grating acts as a Quasi Phase Matching element to couple the light from the laser waveguide to the fiber-like glass waveguide.
The optical and electronic properties of a semiconductor depend on the crystal structure of the device, which has led to investigative work into fabricating artificial structures or superlattices. By tailoring the crystal structure of a device during its fabrication, the optical and electronic properties can be controlled. The crystal structures of such devices may be controlled, for instance, by molecular beam epitaxy (MBE) and metal-organic chemical vapor deposition (MOCVD). Such techniques are capable of monolayer control (xcx9c5 angstrom) over the chemical composition of a crystal.
Other commonly used heterostructures are quantum wells, in which a single layer of one semiconductor is sandwiched between two layers of a larger bandgap material. Strain is produced by using an epitaxial layer with a different lattice constant than the substrate. This strain has a dramatic effect on the properties of the optical system. Among other things, it can allow bandgap tunability, reduced threshold current, and improved laser reliability.
Strain can also allow laser emission to have tailored polarization. By using appropriate strain, one can produce light predominantly polarized as TE, or TM.
Grating-Outcoupled Surface-Emitting Lasers With Flared Gain Regions
This application discloses an innovative laser system. In a preferred embodiment, the laser system comprises a cavity with reflectors on both ends. There is an outcoupling aperture between the reflectors to couple light out of the surface of the device. In a preferred embodiment, there are gain regions located on either side of the outcoupling grating, between the reflectors. The gain regions increase in width as they approach the outcoupling aperture, where they are at their maximum. The gain region can also have straight, non-tapering gain regions located farthest from the outcoupling aperture. The gain regions can have varying contacts as well, which pump the active regions with varying threshold currents depending on the width of the active region at that location.
Our approach avoids the problems cited in OCG devices because it shows innovative grating emitter structures located within the laser cavity and independent of the type of reflectors, these allow efficient laser operation. We teach how to shape the area and vary the properties of the grating emitters to produce desired output beams and help stabilize the laser mode to enhance spatial and temporal coherence. We also show that embodiments that produce several wavelengths for efficient coupling and multiplexing to broad band optical fibers, our structures also allow integration with many devices including broad band modulators, switches, and isolators.
This design makes it easier to limit the length of the outcoupling grating to short lengths, on the order of about 10 microns. In designs where the outcoupler is outside the DBR laser region, it becomes very difficult to outcouple 100% of the light in such a short distance. The light that is not coupled out is wasted and decreases device efficiency.
The disclosed innovations, in various embodiments, provide one or more of at least the following advantages:
low cost;
device testing at the wafer level;
emission at all wavelengths from 0.6-2.0 microns with existing and common material systems, with greater ranges possible;
emission is easily extended to any wavelength as new material systems mature and/or are developed;
low drive currents;
higher power capability than existing VCSELs;
high efficiency;
direct replacement for VCSELs;
easily coupled to multi-mode and single-mode fibers.